Railway: 4.8 Million Users Urged to Rotate Secrets After Environment Variable Breach

Railway, a Platform-as-a-Service cloud provider with roughly 4.8 million users, disclosed on Wednesday, October 22, 2026, that an unauthorized third party may have accessed customer environment variables on its systems and, in some cases, obtained their decrypted contents, in what the company describes as a suspected campaign targeting AI inference providers. "Our investigation to date indicates that an unauthorized third party may have accessed encrypted environment variables and, for some of them, obtained the decrypted contents," the company said, according to The Hacker News. Environment variables on the platform can hold API keys, database credentials, and private registry credentials, and Railway says the impact is limited to a "subset of users" whose exact number it has not disclosed. All users are urged to rotate their secrets.

What was the specific vulnerability?

The exposure centered on the systems Railway uses to store customer environment variables, where an unauthorized third party may have reached both encrypted values and, for some of them, the decrypted contents. According to BleepingComputer, those variables may contain API keys, database credentials, and private registry credentials, meaning a successful read could hand an attacker working access to customer databases, third-party services, and container registries.

Railway says the incident is likely part of a series of attacks aimed at platform companies with the ultimate goal of compromising AI inference providers, the services that run AI models and return answers to Railway-hosted applications. The company notified upstream AI inference providers directly and advised them to revoke all of their keys as a precaution. Separately, security researcher Taylor Monahan (@tayvano_) reported that cryptocurrency wallets were drained using private keys that had been stored on Railway, attributing the theft to attacker address 0xd9773325282874d18cb83f1fe50f14c39750b2ee. The first known crypto theft transaction occurred at 7:36 p.m. on October 22, at least two hours before Railway's official disclosure.

How was the incident detected?

Railway detected suspicious activity on the systems that store customer environment variables earlier in the week of October 20, 2026, and disclosed the incident publicly on Wednesday, October 22, 2026. The company has not published a detailed timeline of how the anomalous activity was first flagged, but its disclosure states that the investigation to date points to unauthorized access to encrypted environment variables and, for some of them, the decrypted contents.

Independent detection ran ahead of the official statement: crypto theft researcher Taylor Monahan traced the first theft transaction to 7:36 p.m. on October 22, at least two hours before Railway disclosed the incident. The company says it will directly notify users whose accounts are confirmed compromised.

What is the current impact?

The confirmed impact covers a "subset of users" whose environment variables may have been read, with cryptocurrency wallets whose private keys were stored on Railway drained in the meantime. Railway hosts more than 16 million services for its roughly 4.8 million users, so the operational fallout extended well beyond the affected subset: the company paused all deployments during the investigation, rotated all Railway bucket credentials, and invalidated Railway API tokens.

As of October 22, 08:44 UTC, builds had been restored for Pro and Enterprise workspaces. Builds for Hobby, Free, and Trial workspaces had not yet resumed, and many deployments remained queued with longer-than-usual wait times. Upstream AI inference providers were advised to revoke all of their keys as a precaution, and all Railway users were urged to rotate secrets stored on the platform.

The incident also lands in a broader context of cloud-credential attacks on AI workloads. As The Hacker News reports, Wiz researchers Shira Ayal and Andre Maccarone documented attackers harvesting GCP credentials via sponsored Google Search ads and running AI workloads in attacker-controlled projects: "Because the AI workloads were processed in foreign projects, prompt contents and model outputs remained completely hidden from the victim organization." Symantec's Threat Hunter Team separately documented an offensive AI agent operating in "Skynet mode" that hijacked exposed LLM endpoints to scan, exploit, steal credentials, and mine crypto for up to 100,000 autonomous steps. Dick O'Brien, principal intelligence analyst at Symantec, told The Hacker News there is no evidence linking that activity to the Railway incident, but noted: "API keys for AI models are valuable because anyone holding one can run inference at the owner's expense."

What are the next steps?

Railway is asking every user to rotate environment variables stored on the platform, including API keys and private registry credentials, and to review service logs for unusual activity. The company has already rotated all Railway bucket credentials, invalidated Railway API tokens, and paused deployments while the investigation proceeds, with builds restored for Pro and Enterprise workspaces as of October 22, 08:44 UTC and Hobby, Free, and Trial builds still down at the time of reporting.

Users whose accounts are confirmed compromised will be notified directly by Railway. Upstream AI inference providers have been advised to revoke all of their keys as a precaution. The company has not disclosed how many accounts were affected or how the unauthorized access was obtained, and the investigation remains ongoing.

By the numbers

• Railway users: approximately 4,800,000
• Services hosted on Railway: 16,000,000+
• First known crypto theft transaction: 7:36 p.m., October 22, 2026, at least two hours before official disclosure
• Builds restored for Pro and Enterprise workspaces: October 22, 2026, 08:44 UTC
• Symantec "Skynet mode" agent autonomous steps: up to 100,000

Source snapshot